EasyPR Versions Save

(CGCSTCD'2017) An easy, flexible, and accurate plate recognition project for Chinese licenses in unconstrained situations. CGCSTCD = China Graduate Contest on Smart-city Technology and Creative Design

v1.5beta

7 years ago
  1. Add a new plate detection method based on text detection (MSER), which is robust to low contrast, less light, and large image.
  2. A new plate detection and recognition evaluation protocol, the detection protocol is similar as ICDAR 2003 evaluation protocol. 50 new images are added to test set.
  3. Non-maximum suppression method used to remove near plates, which results in a reason output of the results. only one images will be output when two methods detect plate in one area.
  4. Spatial-ostu and adaptive threshold plus small sliding window search method used to improve the character segmentation and recognition accuracy.
  5. New SVM model and features (use LBP as it). New ANN model which trained only for Chinese, improve the Chinese recognition accuracy.
  6. Add grid search method to fine tuning the parameters.
  7. Support for OpenMP, the speed using multi thread could be 2 times faster than single thread.
  8. Remove some Chinese to improvement the compatible of the final code in visual studio.

many other improvements and changes, could be found in source code......